Marketing is waiting again. The audience analysis we need sits behind a customer migration and a regulatory deadline, and however useful it might be for our next campaign, I can see why it isn’t first in the queue. By the time it arrives, the opportunity may have passed.
We can argue for a higher priority. Sometimes we should. But when the same sort of request keeps coming back, I start to wonder whether we are asking our colleagues to do work we could handle ourselves, given some help at the beginning.
A small example in McKinsey’s conversation with Slack CMO Ryan Gavincaught my attention for that reason. A colleague working on advertising optimisation had relied on data scientists to supply cohorts for performance analysis. He built a data-science agent to do much of the analysis, with people validating the output. Gavin calls these colleagues “marketing makers”.
I like the possibility this opens up. Someone who understands the campaign can explore a question, see something unexpected, and follow it without having to write another brief and wait for another slot.
Focus On: What could marketing do for itself?
Low-code and no-code tools could put more of that work within reach. We have some evidence that AI can help people work beyond their specialisms: in a study of 776 P&G professionals tackling product-innovation challenges, commercial and R&D participants using AI produced proposals that drew more evenly on both fields. That makes me curious about how much marketing could take on. It doesn’t tell us who should run the resulting tools, or what data they should be allowed to use.
Those questions need answering before an enthusiastic team quietly builds its own little IT department. I doubt our colleagues would thank us for that, particularly when something breaks.
My starting point would be access to datasets, low/no-code capabilities, and pre-agreed approvals. Agree them together. Marketing needs to know what it can use and do without reopening the discussion every time someone wants to try a different analysis.
Following the question
Suppose we want to understand why one audience responds better to an offer than another. The dataset is approved for that purpose, the customer definitions are agreed, and someone in the team can check the analysis. Let them explore it, with marketing responsible for the campaign decisions that follow. A different filter or a follow-up question within that scope should not mean going back to the end of the queue.
Then someone suggests bringing in personal information from another system. Perhaps it would explain the difference; perhaps the team also discovers that its definition of a customer isn’t the one used elsewhere in the business. Now we need the data owner, with privacy or legal colleagues where appropriate. The marketer still has plenty to contribute, including explaining why the question matters, but being able to join two tables does not settle whether we should.
The request may change again. Having found a promising audience, the team wants its agent to update records in the live CRM. I would involve IT before giving a newly built tool those credentials, even if the prototype is working beautifully. The same goes for a shared data pipeline or an integration that other teams rely on.
There is a difference between being able to build something and being equipped to look after it. Who will maintain it when the person who made it moves on? If the answer is a surprised colleague in IT, we have more work to do together before it goes live.
Money brings another boundary. I would expect marketers to make campaign choices within their existing spending authority; an agent’s promising result does not give the team authority to increase it. New commitments need finance’s involvement, as do changes to the assumptions used to report sales. And I would rather have a conversation with finance about what the result means than accept a wonderfully precise ROI figure that nobody there recognises.
Working through one example like this should leave the team with permissions it can actually use, and people to call when the question changes. Some requests will still belong with specialists. If we cannot check the output, or maintaining a tool takes more effort than asking for help, there is little point in insisting on doing it ourselves.
People also need time to learn. “Maker” cannot simply become another responsibility added to an already full marketing role.
Does it get us to market sooner?
I would begin with one recurring workflow that has been held up by competing priorities, and agree the access and approvals with the colleagues responsible. Then try it on the next piece of work. If every instance still needs the same round of permissions, we haven’t changed much.
Time to market is the first measure I would watch, counting the whole journey from an agreed question to a live experiment, including checking and rework. A prototype finished on Tuesday is of limited comfort if the campaign still waits until next month.
Sales growth takes longer to establish. For a targeting change, I would ask finance to help compare the new approach with business as usual, using a suitable customer or account holdout and comparable budgets. We would need enough customers for a meaningful comparison and enough time for the sales cycle to play out. Build, review, and maintenance costs belong in that assessment too. Faster delivery is worth having; it is not evidence of additional sales on its own.
There is a conversation here that I would welcome with IT and data: which request do you keep getting from us, and what would it take for us to handle it properly? I suspect we would learn quite a lot from the answer, including where marketing needs to invest in its own people.
Follow me
That’s all for this week. To keep up with the latest in generative AI and its relevance to your digital transformation programs, follow me on LinkedIn or subscribe to this newsletter.
Disclaimer: The views and opinions expressed in Chronicles of Change and on my social media accounts are my own and do not necessarily reflect the official policy or position of S&P Global.
